Limitation of VaR

A broadly recognized
criterion is the merit of VaR. Once VaR is accepted by the regulators as a main
risk administration tackle. It will turn into a united role, and offers a well
foundation towards the financial institutions, financial intermediaries,
portfolios contraction (Wiener, 1997). In addition, VaR methods: VCV, HS, MCS are
the best methods to examine historical figures and measure the uncomplicated
portfolio, if no other assesses options at linear portfolio (?orkalo, 2011).


Nonetheless, each VaR
methods: VCV, HS, MCS also have their own limitation, they can be erroneous.


Firstly, many VaR has
tail risk, which may understate the amount of VaR, the use of differ data could
result with differ worth. Even if they try to measure the same portfolio, the
results will be different (HKMA; Yamai & Yoshiba, 2002).


Secondly, both the HS
and VCV depend heavily on historical data to predict future price, future may
not replicate the past and the past prediction can be incorrect, especially
under a stress situation, whereas some aberrant volatility arises during the
observation time could result untrue consequents (?orkalo, 2011; HKMA).


Third, historical
correlations are not stable. Although, the historical correlations are presumed
as stable, among market factors and financial instrument, the volatility and
correlations be able to change with time, especially under the stress periods
(?orkalo, 2011; HKMA).


Also, since VaR
considering the losses of market risk and other risk (political risk, liquidity
risk), an extremely cramped centre focal point can be existed (?orkalo, 2011).


Lastly, VaR require a
very large data base, and it is time consuming and the computationally
intensive. To handle the calculation and updating, there will be a great demand
on the computer system capability and human resources (HKMA).




Ongoing Development

In the past days, VaR has become a
common risk measure, but its limitation cannot be ignored. In order to solve
and alleviate the above limitation of VaR, we will next propose the latest VaR
measures, Conditional value at risk
(CVaR), and explain whether the relevant measures will overcome the above


Conditional VaR

Since investors lost a
lot of money in the global financial crisis, after that, conditional value at
risk (CVaR) have been adopted. CVaR is also known as “tail VaR” and “expected
tail loss.” CVaR is an extension of VaR, It devised to estimate the risk of
exceeding losses (Lleo, 2010). A CVaR forecast must be greater than a VaR
forecast, which is computed by “taking a weighted average of the VaR
estimate and the expected losses beyond VaR” (Kidd, 2012, p.2).


The connection
between VaR and CVaR are as follows (Lleo, 2010, p.86):


CVaR is better than
VaR, since it quantifies tail risk and the sub additive was demonstrated, and
it can be able to catch the minimum probability loss for an unsymmetrical risk
(Kidd, 2012). Also, it is a nonparametric tool, which highly improve the
accuracy of the risk prediction, and can solve the problem of accuracy of VaR
(Rockafellar & Stanislav, 2000).


However, CVaR also has
its disadvantage, like other risk methods. With the same confidence level,
using VaR is better than the CVaR, since the stability, and unlike CVaR, VaR do
not need to observe many details before the perdition (Yamai and Yoshiba 2002).